Jupiter SighHi Fedi, I'm looking for a post from a while back that linked to a write up a blind woman did of her assessment of the usefulness of AI/LLMs for blind people. Does anyone remember it? Unfortunately I didn't bookmark it.
Thank you!
Legends and Lotties@jupitersigh I put lots of post about this subject. What was it about specifically?
@Lottie It was a woman (I think, from memory) talking about her personal experience and assessment of the usefulness of AI/LLMs for blind people. My take away from it was that the biggest benefit it offered her was that she didn't need to ask a person for help and didn't have to pay any of the emotional labour involved in doing that. But the downside lay in how the AIs she tested would be confidently incorrect in the same way they would be confidently correct. So she concluded that it was useful tech for her for situations where her personal safety wasn't an issue, like deciding what to wear. But that it wasn't useful when her safety was in question, like accurately reading medication labels. Does any of that sound familiar to you? I remember the post was a link to somewhere else with the full write up.
@jupitersigh I did develop and write about the three layers of trust human in the loop mixture models and when it’s just information. And I did also develop the mixture of models method used in PiccyBot. Do you by any chance write extremely long descriptions? Because if you do, I sent remember talking to you many times.
@Lottie No, that wasn't me, sorry. The post I'm looking for was someone's personal experience, it wasn't about developing things.
@jupitersigh No worries, I think the person I was thinking of was called Jupiter Rowland.